Automatic file organization within a cloud storage system

Abstract

Techniques are described herein for enabling more computationally efficient organization of files within a cloud storage system. A method includes: receiving information identifying a document and a set of folders; for each folder in the set of folders, using a trained model to predict a similarity measure between the folder and the document; for each folder in the set of folders, determining a score for the folder based on the predicted similarity measure for the folder; selecting a candidate folder from the set of folders using the scores of the folders within the set of folders; and providing, on a user interface, a selectable option to associate the document with the candidate folder.

What is claimed is: 
     
       1. A method implemented by one or more processors, the method comprising:
 receiving information identifying a document and a set of folders; 
 for each folder in the set of folders, using a trained model to predict a similarity measure between the folder and the document, wherein using the trained model to predict the similarity measure for each folder comprises:
 processing, using the trained model, one or more folder features of the folder along with one or more document features of the document; and 
 generating the similarity measure for the folder based on the processing; 
 
 for each folder in the set of folders, determining a score for the folder based on the predicted similarity measure for the folder and a folder weight, wherein the folder weight is based on a frequency of access for the folder or a number of files in the folder; 
 selecting a candidate folder from the set of folders using the scores of the folders within the set of folders; 
 providing, on a user interface, a selectable option to associate the document with the candidate folder; 
 receiving an indication of acceptance of the selectable option to associate the document with the candidate folder; and 
 in response to (i) providing the selectable option to associate the document with the candidate folder and (ii) receiving the indication of acceptance of the selectable option to associate the document with the candidate folder, labeling the document with a training label based on the indication of acceptance of the selectable option to associate the document with the candidate folder, and using the document labeled with the training label to further train the trained model. 
 
     
     
       2. The method according to  claim 1 , further comprising:
 in response to receiving the indication of acceptance of the selectable option to associate the document with the candidate folder, automatically associating the document with the candidate folder. 
 
     
     
       3. The method according to  claim 2 , wherein automatically associating the document with the candidate folder comprises moving the document into the candidate folder. 
     
     
       4. The method according to  claim 2 , wherein automatically associating the document with the candidate folder comprises applying a label to the document based on the candidate folder. 
     
     
       5. The method according to  claim 1 , wherein the document and the set of folders are stored on a cloud storage system. 
     
     
       6. The method according to  claim 1 , wherein processing, using the trained model, the one or more folder features of the folder along with the one or more document features of the document comprises:
 determining a vector representation of the document; and 
 determining a vector representation of the folder, and 
 wherein generating the similarity measure for the folder based on the processing comprises determining a similarity between the vector representation of the document and the vector representation of the folder. 
 
     
     
       7. The method according to  claim 6 , wherein the vector representation of the document and the vector representation of the folder have a same dimensionality. 
     
     
       8. The method according to  claim 6 , wherein determining the similarity between the vector representation of the document and the vector representation of the folder comprises determining a cosine similarity. 
     
     
       9. The method according to  claim 1 , wherein selecting the candidate folder from the set of folders using the scores of the folders within the set of folders comprises selecting the candidate folder based on the score for the candidate folder satisfying a threshold. 
     
     
       10. The method according to  claim 1 , further comprising:
 selecting at least one additional candidate folder from the set of folders based on, for each of the at least one additional candidate folder, the score for the additional candidate folder satisfying the threshold; and 
 providing, on the user interface, for each of the at least one additional candidate folder, a selectable option to associate the document with the additional candidate folder. 
 
     
     
       11. The method according to  claim 1 , further comprising:
 determining, based on the scores for the folders within the set of folders, an additional candidate folder; and 
 avoiding providing, on the user interface, a selectable option to associate the document with the additional candidate folder based on the score for the additional candidate folder not satisfying a threshold. 
 
     
     
       12. The method according to  claim 1 , wherein providing, on the user interface, the selectable option to associate the document with the candidate folder comprises:
 in response to the score for the candidate folder satisfying a first threshold and satisfying a second threshold, automatically displaying the selectable option to associate the document with the candidate folder; and 
 in response to the score for the candidate folder satisfying the first threshold but not satisfying the second threshold, only displaying the selectable option to associate the document with the candidate folder subsequent to receiving, via the user interface, a user input that is a request to display the selectable option to associate the document with the candidate folder. 
 
     
     
       13. The method according to  claim 1 , wherein providing, on the user interface, the selectable option to associate the document with the candidate folder comprises:
 providing, on the user interface, an indication that an organization suggestion for the document is available; and 
 in response to receiving, via the user interface, a user input that is associated with the indication that the organization suggestion for the document is available, providing the selectable option to associate the document with the candidate folder, wherein the selectable option indicates a name of the candidate folder. 
 
     
     
       14. A computer program product comprising one or more non-transitory computer-readable storage media having program instructions collectively stored on the one or more non-transitory computer-readable storage media, the program instructions executable to:
 receive information identifying a folder and a set of documents; 
 for each document in the set of documents, use a trained model to predict a similarity measure between the document and the folder, wherein using the trained model to predict the similarity measure for each document comprises:
 processing, using the trained model, one or more document features of the document along with one or more folder features of the folder; and 
 generating the similarity measure for the document based on the processing; 
 
 for each document in the set of documents, determine a score for the document based on the predicted similarity measure for the document and a folder weight, wherein the folder weight is based on a frequency of access for the folder or a number of files in the folder; 
 select a candidate document from the set of documents using the scores of the documents within the set of documents; 
 provide, on a user interface, a selectable option to associate the candidate document with the folder; 
 receive an indication of acceptance of the selectable option to associate the candidate document with the folder; and 
 in response to (i) providing the selectable option to associate the candidate document with the folder and (ii) receiving the indication of acceptance of the selectable option to associate the candidate document with the folder, labeling the document with a training label based on the indication of acceptance of the selectable option to associate the candidate document with the folder, and using the document labeled with the training label to further train the trained model. 
 
     
     
       15. The computer program product according to  claim 14 , the program instructions further being executable to:
 in response to receiving the indication of acceptance of the selectable option to associate the candidate document with the folder, automatically associate the candidate document with the folder. 
 
     
     
       16. The computer program product according to  claim 15 , wherein automatically associating the candidate document with the folder comprises moving the candidate document into the folder. 
     
     
       17. The computer program product according to  claim 15 , wherein automatically associating the candidate document with the folder comprises applying a label to the candidate document based on the folder. 
     
     
       18. A system comprising:
 a processor, a computer-readable memory, one or more computer-readable storage media, and program instructions collectively stored on the one or more computer-readable storage media, the program instructions executable to: 
 receive information identifying a document and a set of folders; 
 for each folder in the set of folders, use a trained model to predict a similarity measure between the folder and the document, wherein using the trained model to predict the similarity measure for each folder comprises:
 processing, using the trained model, one or more folder features of the folder along with one or more document features of the document; and 
 generating the similarity measure for the folder based on the processing; 
 
 for each folder in the set of folders, determine a score for the folder based on the predicted similarity measure for the folder and a folder weight, wherein the folder weight is based on a frequency of access for the folder or a number of files in the folder; 
 select a candidate folder from the set of folders using the scores of the folders within the set of folders; 
 provide, on a user interface, a selectable option to associate the document with the candidate folder; 
 receive an indication of acceptance of the selectable option to associate the document with the candidate folder; and 
 in response to (i) providing the selectable option to associate the document with the candidate folder and (ii) receiving the indication of acceptance of the selectable option to associate the document with the candidate folder, labeling the document with a training label based on the indication of acceptance of the selectable option to associate the document with the candidate folder, and using the document labeled with the training label to further train the trained model. 
 
     
     
       19. The method according to  claim 1 , wherein, for each folder in the set of folders, the folder weight is further based on a recency of access for the folder.